How Should We Measure Social Deprivation in Orthopaedic Patients?

BACKGROUND: Social deprivation negatively affects a myriad of physical and behavioral health outcomes. Several measures of social deprivation exist, but it is unclear which measure is best suited to describe patients with orthopaedic conditions. QUESTIONS/PURPOSES: (1) Which measure of social deprivation, defined as "limited access to society's resources due to poverty, discrimination, or other disadvantage," is most strongly and consistently correlated with patient-reported physical and behavioral health in patients with orthopaedic conditions? (2) Compared with the use of a single measure alone, how much more variability in patient-reported health does the simultaneous use of multiple social deprivation measures capture? METHODS: Between 2015 and 2017, a total of 79,818 new patient evaluations occurred within the orthopaedic department of a single, large, urban, tertiary-care academic center. Over that period, standardized collection of patient-reported health measures (as described by the Patient-reported Outcomes Measurement Information System [PROMIS]) was implemented in a staged fashion throughout the department. We excluded the 25% (19,926) of patient encounters that did not have associated PROMIS measures reported, which left 75% (59,892) of patient encounters available for analysis in this cross-sectional study of existing medical records. Five markers of social deprivation were collected for each patient: national and state Area Deprivation Index, Medically Underserved Area Status, Rural-Urban Commuting Area code, and insurance classification (private, Medicare, Medicaid, or other). Patient-reported physical and behavioral health was measured via PROMIS computer adaptive test domains, which patients completed as part of standard care before being evaluated by a provider. Adults completed the PROMIS Physical Function version 1.2 or version 2.0, Pain Interference version 1.1, Anxiety version 1.0, and Depression version 1.0. Children ages 5 to 17 years completed the PROMIS Pediatric Mobility version 1.0 or version 2.0, Pain Interference version 1.0 or version 2.0, Upper Extremity version 1.0, and Peer Relationships version 1.0. Age-adjusted partial Pearson correlation coefficients were determined for each social deprivation measure and PROMIS domain. Coefficients of at least 0.1 were considered clinically meaningful for this purpose. Additionally, to determine the percentage of PROMIS score variability that could be attributed to each social deprivation measure, an age-adjusted hierarchical regression analysis was performed for each PROMIS domain, in which social deprivation measures were sequentially added as independent variables. The model coefficients of determination (r2) were compared as social deprivation measures were incrementally added. Improvement of the r2 by at least 10% was considered clinically meaningful. RESULTS: Insurance classification was the social deprivation measure with the largest (absolute value) age-adjusted correlation coefficient for all adult and pediatric PROMIS physical and behavioral health domains (adults: correlation coefficient 0.40 to 0.43 [95% CI 0.39 to 0.44]; pediatrics: correlation coefficient 0.10 to 0.19 [95% CI 0.08 to 0.21]), followed by national Area Deprivation Index (adults: correlation coefficient 0.18 to 0.22 [95% CI 0.17 to 0.23]; pediatrics: correlation coefficient 0.08 to 0.15 [95% CI 0.06 to 0.17]), followed closely by state Area Deprivation Index. The Medically Underserved Area Status and Rural-Urban Commuting Area code each had correlation coefficients of 0.1 or larger for some PROMIS domains but neither had consistently stronger correlation coefficients than the other. Except for the PROMIS Pediatric Upper Extremity domain, consideration of insurance classification and the national Area Deprivation Index together explained more of the variation in age-adjusted PROMIS scores than the use of insurance classification alone (adults: r2 improvement 32% to 189% [95% CI 0.02 to 0.04]; pediatrics: r2 improvement 56% to 110% [95% CI 0.01 to 0.02]). The addition of the Medically Underserved Area Status, Rural-Urban Commuting Area code, and/or state Area Deprivation Index did not further improve the r2 for any of the PROMIS domains. CONCLUSION: To capture the most variability due to social deprivation in orthopaedic patients' self-reported physical and behavioral health, insurance classification (categorized as private, Medicare, Medicaid, or other) and national Area Deprivation Index should be included in statistical analyses. If only one measure of social deprivation is preferred, insurance classification or national Area Deprivation Index are reasonable options. Insurance classification may be more readily available, but the national Area Deprivation Index stratifies patients across a wider distribution of values. When conducting clinical outcomes research with social deprivation as a relevant covariate, we encourage researchers to consider accounting for insurance classification and/or national Area Deprivation Index, both of which are freely available and can be obtained from data that are typically collected during routine clinical care. LEVEL OF EVIDENCE: Level III, therapeutic study.

Increasing Rate of Pulmonary Embolism in Trauma Patients at a Level One Trauma Center: An Opportunity for Quality Improvement?

Introduction Pulmonary embolism (PE) is the most common cause of preventable hospital death in trauma patients, with 100,000 patients dying from PE annually. A steadily increasing PE rate was observed over seven years in the trauma population at a single level one trauma center. Our study seeks to analyze this trend by examining risk factors and searching for targets for improvement. We hypothesized that a change in one or more modifiable risk factors was associated with the increased PE rate.  Methods This retrospective cohort study considered trauma patients admitted to our trauma center between 2012 and 2018. The change in PE rate over time and correlation with various risk factors were examined using logistic regression. The study population was divided into two cohorts: early (2012-2015), and late (2016-2018). Data were collected from a prospectively maintained trauma database. More detailed information was obtained from individual patient charts for 533 patients worked up for PE. Risk factors were evaluated using both univariate and multivariate analysis. Results A total of 14,986 trauma patients were included in the study, of which 132 were diagnosed with PE. The PE rate was 1.11% in the late group compared to 0.67% in the early group (p=.004). We detected no association between the PE rate and preventive measures such as screening for and treating deep venous thrombosis (DVT), placing inferior vena cava (IVC) filters, and patterns of chemical DVT prophylaxis. We did not observe a distal migration of the anatomic distribution of PEs on CT pulmonary angiogram (CTPA). There were nonsignificant trends between PE rate and changes in population demographics and injury patterns, increased frequency of major surgery, and increased tranexamic acid (TXA) use. Of known risk factors for PE, units of packed red blood cells (PRBC) (p=0.041), units of fresh frozen plasma (FFP) (p=.037), and the number of patients receiving transfusion (p=0.043) were all significantly greater in the later period. Conclusion Change in hemostatic resuscitation practices (use of balanced ratios of blood products) is most likely to have contributed to the increased PE rate at our institution. However, PE in trauma is multifactorial, and the increased rate cannot be attributed to any single factor. We did not observe a lapse in preventive measures commonly considered indices of quality of care. Caution is advised against overreliance on PE rate as a measure of quality.